`VP_INSTRUCTIONS.md`

25 lines
# Marketing/VP worker — WebSearch ($0, no Exa)

You find each company's Marketing VP/lead using the FREE host **WebSearch** tool (never
Exa, never a paid API). Provenance: contacts are Steve Bloom's (RENTV) — you only ADD.

## Steps
1. Read your batch: `~/Projects/rentv-sheet-enrich/data/vpbatches/vp_<NN>.json` → `targets` = list of `{company}`.
2. For EACH company, call **WebSearch** (you may run several in parallel per message):
   query = `<company> chief marketing officer OR VP marketing OR marketing director LinkedIn`,
   `allowed_domains: ["linkedin.com"]`.
3. From the returned Links, pick the SINGLE best `linkedin.com/in/<slug>` result whose TITLE
   names a MARKETING role (Chief Marketing Officer / VP Marketing / Marketing Director /
   Head of Marketing / Marketing Manager / Communications Director) **at THIS company**.
   - Title format is usually `Name - Title` or `Name - Title at Company`. Extract name + title.
   - IGNORE: /jobs/ links, /company/ links, job postings, and people clearly at a DIFFERENT
     company. If no marketing person clearly ties to the company, leave it blank.
4. **HARD RULE — no fabrication.** Only use a `linkedin.com/in/` URL that literally appears in
   the WebSearch results. Never invent a name or URL. A blank + a "why" beats a guess.
5. Build a JSON list of `{company, mvp, why}` where:
   - `mvp` = `"Name — Title — https://www.linkedin.com/in/<slug>"`
   - `why` = short rationale (e.g. "CMO — top marketing exec", "no marketing lead surfaced")
   Save to `/tmp/vpresult_<NN>.json`, then run:
   `python3 ~/Projects/rentv-sheet-enrich/write_by_name.py 3823360 /tmp/vpresult_<NN>.json`
6. Return ONE line: `vp batch <NN>: <N> companies, <k> marketing VPs found`.